In the present study a large number of high-quality Pn arrival times were hand-picked from seismograms recorded by the portable seismic stations, deployed from 2010 to 2011 in the Yunnan region by the Institute of Crustal Dynamics, CEA and recorded by Chinese permanent seismic stations around the eastern Tibetan plateau. Combining with the data from the observational bulletins of Chinese provincial seismic networks, we obtained a new model of seismic velocity and anisotropy of the uppermost mantle under the eastern Tibetan plateau. Our results show obvious lateral heterogeneities in the study region. The old basins and stable blocks, such as the Sichuan and Qaidam basins and Lhasa and Alashan blocks, show prominent high-velocity anomalies, while the Qilian-Qinling fold zone and northern Chuan-Dian diamond-shaped block exhibit relatively weaker high-velocity anomalies. A pronounced north-south trending low-velocity anomaly zone exists from the Songpan-Garze block east of the Longriba fault, southward through the Anning-Zemu fault zone to the southern Chuan-Dian diamond-shaped block. In the Sanjiang fold system, Burma-subduction zone and the region southeast of the Sichuan basin, there are three obvious low-velocity anomalies. Most of crustal major earthquakes occur on the edges of high-velocity anomalies or the transition zone from high to low velocities, suggesting that crustal earthquakes may be somewhat related to the mantle structure. The fast direction of Pn propagation around the eastern Himalayan syntaxis rotates in a clockwise manner, suggesting that it is related to the Indo-Asian collision. The fast direction changes sharply across the Longmenshan fault zone from NE under the Songpan-Garze block in the west to EW under the Sichuan basin in the east, implying that the material flow of the Tibetan plateau is divided into the NE and SW braches when it meets the Sichuan basin. To south of 26 degrees N, in the Chuan-Dian region the Pn fast direction is in an approximate N-S direction, which is consistent with the GPS observations but different from SKS splitting results, suggesting the mechanically coupling of the crust and the uppermost mantle but decoupling at certain depths between the uppermost and lithospheric mantle.